Jielin Fu

Entropy quantification approaches have gained considerable attention in engineering applications. However, certain limitations persist, including the strong dependence on parameter selection, limited discriminating power, and low robustness to noise. To alleviate these issues, this paper introduces two novel algorithms for time series analysis: the ensemble improved permutation entropy (EIPE) and multiscale EIPE (MEIPE). Our approaches employ a new symbolization process that considers both permutation relations and amplitude information. Additionally, the ensemble technique is utilized to reduce the dependence on parameter selection. We performed a comprehensive evaluation of the proposed methods using various synthetic and experimental signals. The results illustrate that EIPE is capable of distinguishing white, pink, and brown noise with a smaller number of samples compared to traditional entropy algorithms. Furthermore, EIPE displays the potential to discriminate between regular and non-regular dynamics. Notably, when compared to permutation entropy, weighted permutation entropy, and dispersion entropy, EIPE exhibits superior robustness against noise. In practical applications, such as RR interval data classification, bearing fault diagnosis, marine vessel identification, and electroencephalographic (EEG) signal classification, the proposed methods demonstrate better discriminating power compared to conventional entropy measures. These promising findings validate the effectiveness and potential of the algorithms proposed in this paper.

In recent years, underwater wireless optical communication (UWOC) has become a potential wireless carrier candidate for signal transmission in water mediums such as oceans. Underwater signal transmission is impaired by several challenges such as turbulence, scattering, attenuation, and misalignment. In this paper, we propose an improved-order successive interference cancellation (I-OSIC) algorithm based on partition space–time block coding (STBC) technology to solve the sub-channel correlation enhancement problem, which is caused by the combined effects of turbulence and link misalignment in the underwater optical massive multiple-input multiple-output (massive MIMO) systems. The partition STBC technology can make the encoded symbols orthogonality of space and time resist random fading under turbulence environments and fully use the communication link of the massive MIMO system. Under link misalignment conditions, the receiver detector will receive multiple beams. The proposed I-OSIC algorithm based on partition STBC can precisely track the degree of link misalignment error and reorder receiver signals based on the minimum interference criterion. It can use the channel matrix to estimate the interference magnitude of the link misalignment, and then eliminate the interference successively by demodulating the least interfered signal first. When the link misalignment error is large, the I-OSIC algorithm can provide a signal-to-noise ratio (SNR) gain of about 3 dB and provides the same error performance compared with the successive interference cancellation algorithm based on the received signal power.

The task offloading in space-aerial-ground integrated network (SAGIN) has been envisioned as a challenging issue. In this paper, we investigate a space/aerial-assisted edge computing network architecture considering whether to take advantage of edge server mounted on the unmanned aerial vehicle and satellite for task offloading or not. By optimizing the energy consumption and completion delay, we formulate a NP-hard and non-convex optimization problem to minimize the computation cost, limited by the computation capacity and energy availability constraints. By formulating the problem as a Markov decision process (MDP), we propose a multiagent deep reinforcement learning (MADRL)-based scheme to obtain the optimal task offloading policies considering dynamic computation request and stochastic time-varying channel conditions, while ensuring the quality-of-service requirements. Finally, simulation results demonstrate the task offloading scheme learned from our proposed algorithm that can substantially reduce the average cost as compared to the other three single agent deep reinforcement learning schemes.